Identification and Characterisation of a Novel Tissue Barrier in Tendon Blood Vessels C

European Cells and Materials Vol. 31 2016 (pages 296-311) DOI: 10.22203/eCM.v031a19 ISSN 1473-2262

THE BLOOD-TENDON BARRIER: IDENTIFICATION AND CHARACTERISATION OF A NOVEL TISSUE BARRIER IN TENDON BLOOD VESSELS C. Lehner1,4,*, R. Gehwolf1, J.C. Ek2, S. Korntner1, H. Bauer3, H-C. Bauer1,4, A. Traweger1 and H. Tempfer1

1Institute of Tendon and Bone Regeneration, Paracelsus Medical University, SCI-TReCS, Salzburg, Austria 2Institute for Neuroscience and Physiology, Department of Physiology, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden 3Department of Ecology and Evolution, University of Salzburg, Salzburg, Austria 4University Clinic for Trauma Surgery and Sports Injuries, Salzburg, Austria

Abstract Introduction

Tissue barriers function as “gate keepers” between different Tendons harbour a population of multipotent stem/ compartments (usually blood and tissue) and are formed progenitor cells, which are heterogeneous and could be by specialised membrane-associated proteins, localising identified in both peritendon and tendon proper. These cells to the apicolateral plasma membrane domain of epithelial are described to reside both in the vascular compartment and endothelial cells. By sealing the paracellular space, the and the tendon proper (Lui, 2013; Tempfer et al., 2009). free diffusion of solutes and molecules across epithelia and The niche of these cells is not yet fully understood. The endothelia is impeded. Thereby, tissue barriers contribute finely tuned molecular and cellular composition of this to the establishment and maintenance of a distinct internal niche appears to be crucial for stem cell differentiation and external environment, which is crucial during organ and maintenance (Bi et al., 2007; Katenkamp et al., 1976; development and allows maintenance of an organ- Tempfer et al., 2009; Zhang and Wang, 2013; Zhang et al., specific homeostatic milieu. So far, various epithelial and 2010b). So far, besides oxygen tension and mechanical endothelial tissue barriers have been described, including stimuli, few other niche-determining factors are known the blood-brain barrier, the blood-retina barrier, the including the composition and nano-/micro-structure of blood-testis barrier, the blood-placenta barrier, and the the extracellular matrix (ECM) (Bi et al., 2007; Rui et al., cerebrospinal fluid (CSF)-brain barrier, which are vital for 2011b; Tong et al., 2012; Wang et al., 2008; Zhang and physiological function and any disturbance of these barriers Wang, 2013). can result in severe organ damage or even death. Here, The internal milieu of the niche is severely changed we describe the identification of a novel barrier, located during tissue degeneration, trauma and injury, often resulting in the vascular bed of tendons, which we term the blood- in pathological alterations such as hyperproliferation, tendon barrier (BTB). By using immunohistochemistry, erroneous tendon cell differentiation, and calcification transmission electron microscopy, and tracer studies we (Oliva et al., 2012; Rui et al., 2011a). Therefore, it appears demonstrate the presence of a functional endothelial reasonable to suggest that rapid restoration and stabilisation barrier within tendons restricting the passage of large of the niche following an injury is paramount in initiating blood-borne molecules into the surrounding tendon and supporting the healing process. tissue. We further provide in vitro evidence that the BTB Healthy tendons are sparsely vascularised (Ahmed potentially contributes to the creation of a distinct internal et al., 1998; Hooper et al., 1984; Weidman et al., 1984; tissue environment impacting upon the proliferation and Zhang et al., 1990), particularly sheathed tendons – such differentiation of tendon-resident cells, effects which might as the flexor tendon of the hand – and depend on synovial be fundamental for the onset of tendon pathologies. fluid diffusion to provide nutrition (Hagberg et al., 1992; Jones et al., 2003; Lundborg et al., 1980; Manske and Keywords: Tendon vasculature, blood vessels, endothelial Lesker, 1985; Tempfer and Traweger, 2015). Between cells, barrier, tight junctions, tracer, vascular permeability, densely packed collagen fibre bundles continuous rows serum, tendinopathy. of tenocytes are found perfectly aligned with the collagen structure in direction of the strain (Kannus, 2000; Legerlotz et al., 2014), exhibiting an exceptionally low turnover rate, which may also account for the poor regenerative capacity of tendon tissue (Heinemeier et al., 2013). *Address for correspondence: In contrast, chronically inflamed tendons or tendons Christine Lehner, PhD which are recovering from injury are characterised by Institute of Tendon and Bone Regeneration hypercellularity and hypervascularity, due to increased Paracelsus Medical University – Spinal Cord Injury & proliferation of tenocytes and vascular cells (Andersson et Tissue Regeneration Center Salzburg al., 2011; Astrom and Rausing, 1995). In vitro, an increase A-5020 Salzburg, Austria in proliferation has been observed in isolated tendon Telephone Number: +43 662 2420-80867 cells when cultivated with serum or platelet rich plasma E-mail: [email protected] (Mazzocca et al., 2012; Wang et al., 2012).

www.ecmjournal.org 296 C Lehner et al. The blood-tendon barrier

So far, tendon vasculature has been investigated Proliferation assay in the context of tendon development, tendinopathies To examine the effect of serum on cell proliferation, human and vascularisation of tendon tissue grafts (Fenwick et TDCs were cultured in the presence of pooled human al., 2002). Little attention has been paid to the tendon serum (Sigma, Vienna, Austria, #H6914) at concentrations vasculature as a restrictive system impeding the passive of 10 %, 5 %, 1 % and without serum. Cells were counted transport of solutes and molecules across vessel walls and after 3 and 5 d. For an additional experiment, we cultured thereby controlling the homeostasis of the tendon-specific human tendon cells either with 10 % serum or without niche. Such influx and efflux control systems usually rely serum. Cells cultured without serum were supplemented on a sophisticated network of membrane-spanning proteins with serum on the eighth day. In parallel, serum was (e.g. occludin, claudins and junctional adhesion molecules) withdrawn from cells cultured with serum so far. Again, linked to a vast array of cytoplasmic proteins, which seal cell numbers were counted every second or third day until the paracellular space between neighbouring cells (Bauer cells reached confluence after two weeks. et al., 2014; Bauer et al., 2011). The aim of this study was to investigate the structural, Differentiation experiments functional and molecular characteristics of blood To assess the effect of serum on tendon cell differentiation vessels in human and mouse tendon tissue. By means potential, human TDCs were cultured with and without of transmission electron microscopy (TEM), RT-PCR, serum supplementation for one week. After this period, immunohistochemistry, and in vivo tracer studies we either Adipolife™, Chondrolife™ or Osteolife™ (LM- showed that the vascular lining of tendon vessels serves 0023, Lifeline Cell Technology, Frederick, MD, USA), as a size-selective diffusion restraint, reminiscent of the serum-free adipogenic, chondrogenic or osteogenic media, tissue barrier located in the cerebral microvasculature of respectively, were added to both cell types for two weeks. vertebrates (Abbott et al., 2010). Moreover, we cultivated Subsequently, 1) adipogenic differentiation was assessed tendon-derived cells with increasing serum concentrations, by Oil red O staining and quantitative RT-PCR (qRT- thus mimicking a barrier breakdown in vitro, and examined PCR) determining the expression level of peroxisome the effect on proliferation and differentiation. proliferator-activated receptor gamma (PPARγ), a regulator We propose that this blood-tendon barrier (BTB) is of adipocyte differentiation (Rosen and Spiegelman, 2001), imperative for the establishment of stable homeostatic 2) chondrogenic differentiation by Alcian blue, collagen conditions, thus potentially contributing to the regulation II and aggrecan staining, 3) osteogenic differentiation of tendon cell differentiation. by staining for alkaline phosphatase, qPCR on three osteogenesis markers (runt-related transcription factor 2, RUNX2; bone sialoprotein 1, BSP1; and parathyroid Materials and Methods hormone 1 receptor, PTH1R) and visualisation of calcium deposits by Alizarin S (Sigma, Vienna, Austria) staining. Sample collection In total, 14 human tendon samples (biceps tendons: ß-Galactosidase staining n = 9, 8 men, 1 woman; mean age 60.4 years, range 30 In order to exclude that lack of serum affects cell to 75, obtained from rotator cuff repairs; semitendinosus senescence, we performed a ß-galactosidase staining tendons: n = 3, 2 men, 1 woman; mean age 31.6 years, according to the protocol described by Dimri et al. (1995). range 18 to 55, obtained from anterior cruciate ligament To this end, cells were cultivated for 2 weeks with either reconstruction; and Palmaris longus tendons: n = 2, 1 Adipolife™ without serum, Adipolife™ with 10 % human man, 1 woman; mean age 56 years, range 37 to 75) were serum, MEM with human serum, Osteolife™ without obtained with patients’ informed consents as approved by serum or Osteolife™ with 10 % human serum. The cells the Salzburg ethics committee (approval #E-Nr. 1809). were fixed for 1 min with 3 % paraformaldehyde (PFA) 15 female C57BL/6 mice (twelve months old) were at room temperature and subsequently washed 3 times provided by Charles River Laboratories (Sulzfeld, with cold phosphate-buffered saline (PBS). The cells were Germany). Care of the animals and all animal experiments then incubated at 37 °C for 8 h with a buffer containing have been conducted according to national and international 40 mM citric acid, 5 mM potassium ferrocyanide, 5 mM guidelines and were approved by the local ethics committee. potassium ferricyanide, 150 mM NaCl, 2 mM MgCl2 and freshly added X-Gal (1 mg/mL) at pH 6. After the cells Cell culture were mounted with cover slips, they were microscopically Freshly obtained biopsy specimens from human biceps analysed. (n = 7, 6 male, 1 female) and semitendinosus (n = 3, 2 male, 1 female) tendons were cut into small pieces under sterile Alizarin S staining and quantification conditions followed by a 12 h digestion step in Minimum In order to visualise and quantitate the amount of Essential Medium (MEM), supplemented with 15 mg/ calcification, cells cultured in 6-well plates were fixed in mL collagenase II (Gibco, Invitrogen, Lofer, Austria) at 4 % PFA for 20 min at room temperature and stained with

37 °C, 85 % humidity and 5 % CO2. Incubated in MEM a solution of 0.1 % Alizarin S at pH 4.5 for 5 min at room supplemented with 10 % foetal bovine serum (FBS) the temperature. After 3 washing steps in double-distilled resulting cell suspension gave rise to an adherent culture H2O to remove any unbound stain, the bound stain was of tendon-derived cells (TDCs), which was passaged once dissolved by using 800 µL of 10 % (v/v) acetic acid under before the experiments were performed. gentle agitation for 30 min. After vortexing for 30 s, the

www.ecmjournal.org 297

tubes were heated to 85 °C for 10 min and then centrifuged controls. After treatment with Power Vision polyHRP-anti- at 15000×g for 10 min. 500 µL of the supernatant was rabbit IgG (ImmunoLogic, Duiven, Netherlands), sections removed and transferred to new 1.5 mL micro-centrifuge were incubated with 3,3’-diaminobenzidine (DAB) tubes and 200 µL of ammonium hydroxide solution 10 % tetrahydrochloride, Sigma Aldrich, Vienna, Austria), and (v/v) was added to achieve a pH of between 4.1-4.5. counterstained with Novocastra™ Haematoxylin dye Aliquots (150 µL) of each sample were read in triplicate (Leica Biosystems, Newcastle, UK) and mounted in Eukitt at an absorbance of 405 nm on a microplate reader (Tristar (Sigma Aldrich, Vienna, Austria). LB 941, Berthold Technologies, Bad Wildbad, Germany). For double-immunolabelling, mouse Achilles tendons In order to correct for varying cell numbers under different were fixed in 4 % PFA, washed with PBS-sucrose and conditions, OD values were normalised to DNA content, cryosectioned on a Leica CM 1950 microtome. Sections which was determined in parallel cultures. (15 µm) were stained with biotinylated Bandeiraea simplicifolia lectin (BSL, obtained from Sigma Aldrich, Oil red O staining and quantification Vienna, Austria) visualising blood vessels, and antibodies To determine the extent of adipogenic differentiation, specific for ZO-1, occludin, claudin 3, and claudin 5. cells cultured in 6-well plates were fixed in 4 % PFA for Primary antibodies were visualised with secondary 20 min at room temperature and stained with Oil red O antibodies labelled with Alexa Fluor 568 (donkey anti working solution for 15 min at room temperature. After rabbit Alexa Fluor 568, Life Technologies, Vienna, 3 washing steps, bound Oil red O was dissolved in 400 µL Austria), BSL was visualised with Streptavidin, Alexa isopropanol and absorbance was measured at a wavelength Fluor 488 conjugate (Life Technologies, Vienna, Austria) of 540 nm (Tristar LB 941, Berthold Technologies). In and analysed on a LSM710 (Zeiss, Munich, Germany). order to correct for varying cell numbers under different Tracer-injected, paraffin-embedded Achilles tendons conditions, OD values were normalised to DNA content, were sectioned (6 µm) and mounted on Superfrost ultra which was determined in parallel cultures. plus glass slides (Thermo Scientific, Vienna, Austria). Following deparaffinisation and dehydration, sections DNA quantification were incubated in blocking reagent (Roth, Graz, Austria) Cells were lysed in 5 mM Tris-HCl (pH 8.0) by 3 freeze/ for 30 min, incubated with a streptavidin-HRP labelled thaw cycles, followed by sonication for 20 min. DNA antibody (Dako, Glostrup, Denmark, #90397) for 1 h, and was measured in the supernatant of the solutions using a developed with DAB after several washing steps in PBS. fluorescent Quant-iT™ PicoGreen® dsDNA reagent assay kit (Invitrogen, Carlsbad, CA,USA, # P7589) according to Immunocytochemistry the manufacturer’s instructions. In order to better characterise the cells used for the in vitro experiments, human TDCs were fixed with 4 % Tracer injection paraformaldehyde, washed with PBS and incubated To assess the tightness of tendon blood vessels, mice with antibodies directed against tenomodulin (Santa (C57BL/6) were injected with 0.10-0.15 mg/g of 10 kD Cruz, Heidelberg, Germany, sc-98875), pro-collagen1a1 biotin-dextran (BDA) per animal (Molecular Probes; (Abcam, Cambridge, UK, ab64409), CD44, CD29, CD90, Carlsbad, CA, USA, #D-1956) or a 287 D biotinylated CD105, and CD45 (Abcam, Cambridge, UK, ab93758) at Neurobiotin™ tracer (Vector Laboratories, Peterborough, 4 °C for 48 h. Secondary antibodies labelled with Alexa UK, #SP-1120) into the tail vein. The tracer was allowed Fluor 568 and cy5 (donkey anti rabbit Alexa Fluor 568 and to circulate for 15 min before the animals were euthanised donkey anti rat cy5, Life Technologies, Vienna, Austria) by an overdose of sodium thiopental (Abcur, Helsingborg, were used to visualise primary antibodies. Sweden) (0.4 g/kg animal) through an IP injection. Achilles tendons were removed, cut into 1 mm3 cubes, immersion- Laser-capture microdissection and RT-PCR fixed in 4 % PFA at 4 °C overnight, thoroughly washed in To specifically analyse tendon blood vessels, laser-capture PBS and further processed for paraffin embedding. Heart microdissection was performed on paraffin-embedded and brain tissue as controls were processed in parallel. tissue sections of mouse Achilles and human biceps tendons. Parts from the vascular bed and the dense Immunohistochemistry collagenous region of the tendon were dissected using a To identify the presence of tight junction proteins in tendon PALM MicroBeam laser capture microscope from Zeiss blood vessels, human tendon biopsy samples (Palmaris and captured tissue was placed in an RNase-free PCR cap. longus, biceps tendon) were fixed in 4 % PFA in PBS at Extracted total RNA was directly transcribed into cDNA 4 °C overnight. Achilles tendons from C57BL/6 mice were using the High Capacity RNA-to-cDNA Master Mix (Life perfusion-fixed with 4 % PFA in PBS and immersion- Technologies). PCR was performed using the smallest fixed in the same fixative at 4 °C overnight. Tissues were available intron-spanning TaqMan Gene Expression processed for paraffin embedding, sectioned (6 µm), and Assay (with amplicons <100 bp, Life Technologies, stained with antibodies specific for zonula occludens Vienna, Austria) for occludin (assay ID: Hs00170162_ (ZO-1; Invitrogen, Lofer, Austria #61-7300), occludin m1; Mm00500912_m1), claudin 1 (Hs00221623_m1; (Invitrogen, #71-1500), claudin 3 (Abcam, Cambridge, Mm01342184_m1), claudin 3 (assay ID: Hs00265816_s1; UK, ab15102), and claudin 5 (Abcam, ab53765). Mm00515499_s1), claudin 4 (assay ID: Mm00515514_s1), Incubation with primary antibodies was performed at 4 °C claudin 5 (assay ID: Hs01561351_m1; Mm00727012_s1), overnight. Primary antibody was omitted for negative and claudin 12 (assay ID: Hs00273258_s1; Mm01316509_

www.ecmjournal.org 298 C Lehner et al. The blood-tendon barrier m1). Amplification efficiency of the house-keeping gene to the expression of three previously validated endogenous hypoxanthine-guanine phosphoribosyltransferase (HPRT, control genes as described by Vandesompele et al. (2002). assay ID: Hs01003267_m1; Mm01545399_ml) was used The normalised quantities were then determined for the as an internal control. PCR products were separated on an candidate genes scaled against the expression values agarose gel and stained with ethidium bromide. determined for the undifferentiated control (MEM+HS) to generate fold changes in expression. Measurements Electron microscopy from three independent experiments (n = 3) are shown For analysis of tendon blood vessels on the ultrastructural (technical replicate for each sample) and expressed as level, 3 mouse Achilles tendons were perfusion-fixed mean ± standard error of the mean (SEM). with 2 % glutaraldehyde in PBS and immersion-fixed in 2 % glutaraldehyde in 100 mM phosphate buffer (PB) Western blot (pH 7.0) at 4 °C overnight. After several washing steps To determine the expression levels of extracellular in phosphate buffer specimens were post-fixed in 1 % matrix proteins, lysates from human tendon cells were glutaraldehyde and 1 % osmium tetroxide in 50 mM PB prepared as described previously (Traweger et al., 2008). (pH 6.2) at 4 °C. Following extensive rinsing in distilled Protein contents were quantified by using a BCA protein water and incubation in 1 % uranyl acetate, tissues were assay kit (Pierce, Rockford, IL, USA). Equal amounts dehydrated in a graded ethanol series and embedded of protein were subjected to SDS-polyacrylamide gel using the Ultra Bed Low Viscosity Epoxy Kit (Electron electrophoresis (PAGE) using a 7.5 % or 10 % gel. After Microscopy Sciences, Hatfield, PA, USA) according to the blotting, the polyvinylidene difluoride (PVDF) membrane manufacturer’s instructions. was incubated in 5 % non-fat milk powder in Tris-buffered Semithin sections were cut and stained with Azur II saline-Tween (TBS-T). Immunodetection was performed to determine the region of interest. Ultrathin sections (~ using primary antibodies recognising pro-collagen1a1 80 nm) were cut on an ultramicrotome Reichert Ultracut of Col1a1 (Santa Cruz, Heidelberg, Germany, sc-8782), S (Optische Werke C. Reichert, Vienna, Austria), mounted MMP2 (Abcam, Cambridge, UK, ab37150), MMP9 on Formvar-coated copper slit grids and contrast stained (Cedarlane, Ontario, Canada), β-actin (Santa Cruz, sc- with uranyl acetate/lead citrate in an ultrostainer (Leica/ 47778) and secondary HRP-labelled goat anti-rabbit and Reichert, Vienna, Austria). All ultrathin sections from the goat anti-mouse antibodies, respectively (BioRad, Munich, proximal, the middle and the distal part of the tendons Germany). Bands were visualised using the SuperSignal were examined in a transmission electron microscope LEO West Pico Chemiluminescent Substrate from Pierce EM 910 (LEO, Elektronenmikroskopie Ltd, Oberkochen, (Thermo Scientific, Vienna, Austria). All experiments Germany) operating at 80 kV. were repeated at least three times. Band intensities were measured densitometrically and normalised to ß-actin Quantitative RT-PCR expression using the Image Lab Software 5.1 from BioRad. To assess the expression levels of genes associated with differentiation into the adipogenic and osteogenic Zymography lineages, we performed quantitative RT-PCR. To this Matrix metalloproteinase activities were assessed by end, total RNA from human tendon cultures was isolated gelatine zymography in tendon cell cultures. To this end, using TRI Reagent (Sigma-Aldrich, Vienna, Austria) and human TDCs were incubated with increasing serum (0 %, reverse-transcribed with iScript Reverse Transcription 1 %, 5 % and 10 %) concentrations. After 4 d, serum- Supermix for RT-qPCR (BioRad, Munich, Germany, # containing medium was removed, cells were washed 170-8841). Products were amplified using the Taq Man twice with PBS and serum-free medium was added to Gene Expression Master Mix (Thermo Fisher Scientific, the cell cultures. After 10 min the supernatant was taken Applied Biosystems, Vienna, Austria, # 4369016) and and gelatine zymography was performed as described the smallest available PrimeTime qPCR assays from previously (Krizbai et al., 2000). Briefly, cells were lysed IDT (Integrated DNA Technologies, Coralville, IA, in a buffer containing 20 mM Tris-HCl (pH 7.5), 150 mM USA) for PPARγ (ID assay: Hs.PT.58.25464465), NaCl, 1 % Triton X-100, and 0.1 % SDS. Equal amounts RUNX2 (ID assay: Hs.PT.56a.19568141), BSP1 (ID of protein were loaded on 7.5 % polyacrylamide gels assay: Hs.PT.58.20755025), and PTH1R (ID assay: containing gelatine (1 mg/mL). Following electrophoresis,

Hs.PT.58.3437800) using a CFX96 Touch™ Real-Time the gels were given a 30 min wash with ddH2O containing PCR Detection System (BioRad, Munich, Germany). PCR 2.5 % Triton X-100, and then incubated with a substrate was performed applying the following temperature regimen: buffer (50 mM Tris-HCl (pH 8.0) containing 5 mM

2 min 50 °C uracil DNA glycosylase (UDG) incubation; CaCl2) overnight at 37 °C. The gels were then stained 10 min 95 °C enzyme activation; 15 s 95 °C, 1 min, with Coomassie Blue R-250 for 30 min followed by 60°C for 40 cycles. For normalisation the house-keeping several washing steps. Band intensities were measured genes hypoxanthine-guanine phosphoribosyltransferase densitometrically using the Image Lab Software 5.1 from (HPRT1, Hs.PT.39a22214821), ribosomal protein, large, Biorad. P0 (RPLP0, Hs.PT.56.40434846), and TATA box-binding protein (TBP, Hs.PT.39a.22214825), were used as internal Statistical analysis controls. Cq values were analysed using qBasePlus v. 2.4 Statistical analyses were performed using Graph Pad Prism (Biogazelle NV, Zwijnaarde, Belgium) and normalised software (version 5.04). Densitometric data are presented relative quantities were calculated by normalising the data as means with standard deviations. One-way analysis of

www.ecmjournal.org 299

Fig. 1. Representative TEM micrograph of mouse Achilles tendon blood vessel. The endothelium displays a continuous, non-fenestrated endothelium with TJ structures (red arrows). Scale bar: 500 nm. variation (ANOVA) using the nonparametric Kruskal- Immunoreactivity to at least zonula occludens-1 (ZO- Wallis test were used to test for differences between the 1), occludin, claudin 3 and claudin 5 was detected in groups. qPCR data were analysed using the Mann-Whitney endothelial cells lining the blood vessels of human palmaris test. p < 0.05 was considered statistically significant. longus (Fig. 3a-d), human biceps tendon (data not shown) and mouse Achilles tendons (Fig. 3e-h). Double immunolabelling using Bandeiraea simplicifolia Results lectin and antibodies directed against TJ proteins ZO- 1, occludin, claudin 3 and claudin 5 on mouse Achilles In order to characterise a putative BTB in human and cryosections showed that the TJ-proteins, occludin, claudin mouse tendons which regulates the passage of serum 3 and claudin 5 were exclusively localised to the innermost components from the circulation into the tendon tissue, vessel layer, that is the endothelium (Fig. 3e-p). we first set out to demonstrate the presence of structural components known to constitute other tissue barriers. By Tracer studies means of transmission electron microscopy we investigated Since the mere presence of tight junction proteins does the ultrastructure of tendon vessels and found that vessels not necessarily reflect functional tightness of a tissue of both the proximal as well as the distal tendon region barrier, we conducted tracer studies by using both a displayed a continuous, non-fenestrated endothelium biotinylated 10 kD dextran and a small 287 D neurobiotin with a distinct basal lamina. At higher magnification we tracer. Inspection of the streptavidin-HRP-labelled tissues identified “kissing points” indicative for tight-junctional revealed that the 10 kD tracer remained within the vessel complexes (Fig.1). walls of tendon and brain tissue, whereas substantial To demonstrate the presence of tight junctions at the leakage into the heart muscle was evident (Fig. 4a-f). In tendon endothelium we performed a qualitative RT-PCR contrast, the small tracer was only retained by brain vessels, to determine the presence of mRNAs encoding occludin, but passed the endothelium in both the tendon and heart claudins 1, 3, 4, 5, 7, 11, 12 and 19, as well as Tricellulin (Fig. 4g-l). using RNA extracted from laser-microdissected human biceps and mouse Achilles tendon blood vessels (Fig.2). Tendon cell culture With immunohistochemical staining we tested Proliferation assay for several tight junction (TJ)-proteins known to be To better characterise tendon cell cultures obtained present at cerebral and retinal endothelial cell junctions. by collagenase II digestion, we stained for markers

www.ecmjournal.org 300 C Lehner et al. The blood-tendon barrier

Fig. 2. RT-PCR analysis of RNA isolated from human biceps (a, b) and mouse Achilles tendon (c, d) vessels lasermicrodissected from 6 µm thick paraffin sections. mRNAs encoding TJ-associated proteins were analysed. Scale bars: 75 µm.

Fig. 3. Immunohistochemical staining of human palmaris longus (a-d) and mouse Achilles tendons (e-p) stained with antibodies directed against ZO-1 (a, e), occludin (b, f), claudin 3 (c, g) and claudin 5 (d, h). Blood vessels in mouse Achilles tendons are visualised by Bandeiraea simplicifolia lectin (BSL) (i-l), merged images show co-distribution of ZO-1, occludin, claudin 3 and claudin 5 with BSL, and nuclei are stained with DAPI (m-p). Scale bars: 50 µm.

www.ecmjournal.org 301

C Lehner et al. The blood-tendon barrier

Fig. 5. Immunostaining on human biceps tendon cells show the cells coexpressing tenomodulin and pro-collagen1a1 (a-c), CD29 and CD44 (d-f). CD90 (g) and CD105 (h) are expressed, whereas CD45 (i) is not detectable. Scale bars: 100 µm. incubated with serum and MEM, staining for neither analysis of bound Alizarin S shows that addition of collagen II nor aggrecan could be detected (Fig. 8c,g,h). serum causes a significant (p < 0.01) 5.2 fold increase of Only cells exposed to a standard chondrogenic medium mineralisation compared to cultures in unsupplemented stained for Alcian blue (Fig. 8d,e), a dye staining for acidic Osteolife™ (Fig. 9d). Since alkaline phosphatase is not sulphated mucosubstances and acetic mucins, whereas cells exclusively produced by osteoblasts, we performed qPCR in control medium displayed a greenish colour (Fig. 8f). for the osteogenesis markers Runx2, BSP1 and PTH1R Osteogenic differentiation was assessed by Alizarin (Fig. 9k,l,m). Although an increase in expression was S and alkaline phosphatase staining. The serum-free evident for all genes, the difference to the control group osteogenic medium Osteolife™ failed to induce mineral was statistically not significant. deposition and alkaline phosphatase activity in tendon cells (Fig. 9a,e,h), unless supplemented with 10 % human ß-Galactosidase staining serum (Fig. 9b,f,i). In serum-free Osteolife™ treated The number of senescent cells visualised by ß-galactosidase cultures, only few nodules with calcium deposits were staining showed no difference between treatment groups detectable by Alizarin S staining (Fig. 9a). Prominent (data not shown). Under all culture conditions, the number alkaline phosphatase activity could be detected in tendon of positive cells was < 0.1 %. cell cultures supplemented with serum irrespective of the Apart from its influence on proliferation and addition of Osteolife™ (Fig. 9f,g,i,j). Semiquantitative differentiation, serum also affected the expression of the

www.ecmjournal.org 303

Fig. 6. Proliferation of human tendon cells (n = 3 donors) is serum-dependent, as shown by cell counting (a) and MTS assay after 5 d of culture (b). MTS assay shows no difference in cell proliferation between human and bovine serum (FBS). Representative photomicrographs of human tendon cells cultured for 5 d at different serum concentrations illustrate the effect of serum on tendon cell proliferation c( -f), Scale bar: 100 µm. Under serum-free conditions, human tendon cells do not proliferate, addition of serum after 8 d (green arrow) induces proliferation (g, green line). Serum withdrawal (red arrow) in cells cultured for 8 d with 10 % human serum leads to an abatement of proliferation (g, blue line).

matrix degrading matrix metalloproteinases MMP2 and complexes between endothelial cells, similar to vessels in MMP9 and of pro-collagen1a1, the precursor form of the brain and retina (Garbuzova-Davis et al., 2013; Masaki the ECM building protein collagen1. Whereas MMP2 et al., 1995; Wallow and Geldner, 1980). (p = 0.0572) and MMP9 (p = 0.0972) expression increased We further demonstrated the presence of a series of tight markedly with increasing serum concentration, pro- junction molecules, both on the RNA as well as the protein collagen1a1 expression appeared unchanged between 1 level, in human and mouse tendon vessels constituting and 10 % serum (Fig. 10). the prerequisite for the formation of a functional barrier. Interestingly, various core TJ-proteins expressed by tendon endothelial cells (tECs) resemble the set of TJ-proteins Discussion expressed at the blood-brain barrier (BBB) and the blood- retina barrier (BRB). At the TJs of central nervous system In this work, we focus on the structure, composition (CNS)-capillaries, occludin, ZO-1, claudin 3 and claudin and permeability of blood vessels in healthy tendons. 5 have been detected in both human and mouse (Nitta Moreover, we show effects of serum on tendon cells et al., 2003; Schrade et al., 2012; Wolburg et al., 2003). in vitro. Our findings provide evidence to suggest the In rodent retinal endothelial cells, the TJ-proteins ZO-1, existence of a blood-tissue barrier in tendons allowing occludin, claudin1/3, claudin 2, claudin 5 and JAM1 have the establishment of a stable, internal milieu important for been shown to be expressed (Klaassen et al., 2009; Luo et tendon cell proliferation and tendon function. al., 2011; Xu et al., 2005), and claudin 12 transcripts have Since tendons are sparsely vascularised tissues, been detected in bovine retinal endothelial cells (Klaassen ultrastructural analysis has so far mainly focussed on et al., 2009). Claudins 1, 3 and 5 are considered to be fibril size and composition. In our study, therefore, we barrier-sealing proteins (Pfeiffer et al., 2011; Zhang et investigated the ultrastructure of tendon blood vessels. On al., 2010a). In overexpression experiments using Madin- TEM micrographs we found that tendon vessels display a Darby canine kidney (MDCK) cells, Milatz et al. (2010) continuous, non-fenestrated endothelium with junctional showed that claudin 3 acts as a barrier-forming tight

www.ecmjournal.org 304 C Lehner et al. The blood-tendon barrier

Fig. 7. Oil red O staining of human tendon cells cultured with Adipolife™ (a1 overview, a2 photomicrograph), Adipolife™ supplemented with 10 % human serum (b1 overview, b2 photomicrograph), and MEM with 10 % human serum (c1 overview, c2 photomicrograph) reveals that serum promotes the formation of lipid droplets. Scale bar: 100 µm. Semiquantitative analysis of bound Oil red O shows that addition of serum causes a significant (**p < 0.01) 2.5 fold upregulation of lipid droplets (d). Transcripts of PPARγ are significantly (**p < 0.01) upregulated in the cells cultured with Adipolife™ and human serum (e). Scale bars: 150 µm.

junction component that strongly reduces the paracellular on the permeability of tendon vessels exist. Our tracer permeability to ions and larger uncharged solutes but not studies revealed that mouse tendon vessels, in contrast to osmotically driven water. to heart vessels, are tight to a 10 kD tracer, but not to a It is well established that composition and architecture 287 D tracer, suggesting that tendon vessels are not as tight of junctional complexes are critically involved in regulating as brain capillaries which exclude both tracers. The fact vessel permeability (Bazzoni, 2006). Tendon vasculature that the 10 kD tracer does not permeate the blood vessel in general and tendon vessel properties in particular have wall implies that most serum proteins are excluded from hardly been investigated so far. To our knowledge, no data paracellular entry into the tendon tissue. However, we

www.ecmjournal.org 305

Fig. 8. Immunohistochemical staining of pellet cultures from human tendon cells (a-c) show that collagen type II (91.6 % cells positive, g) and aggrecan (76.7 % positive, h) are expressed when the pellets were cultured in unsupplemented Chondrolife™. Upon culture in Chondrolife™ supplemented with 10 % human serum, collagen type II is expressed by 39.6 % of all cells (g) and aggrecan by 3.6 % (h). Neither collagen type II nor aggrecan are detectable when cells were cultured in MEM and serum (g,h) (***p < 0.001). The pellets are positive for Alcian blue staining under all conditions (d-f). Scale bars: 50 µm (a-c), 1 mm (d-f). cannot exclude an active transport via a transporter protein Since the charge- and size-selective properties of the mediating transcellular passage. paracellular barrier are determined by the combination Modulation of vessel permeability by altered TJ and interaction of the claudin family members present composition has clearly been shown by Nitta et al. (2003). within a cell, alterations in the expression of these proteins By performing tracer experiments in claudin 5-deficient have a strong influence on ion homeostasis and, thus, the mice, the authors demonstrated a size-selective increase in microenvironment of the tissue niche (Mizutani et al., the permeability of morphologically normal blood vessels 1995; Wolburg et al., 2003). allowing the passage of small (< 800 D), but not larger In order to obtain information on the effects of serum on molecules. tendon cells, we utilised an in vitro approach to mimic blood

www.ecmjournal.org 306 C Lehner et al. The blood-tendon barrier

Fig. 9. Alizarin S staining of human tendon cells cultured with Osteolife™ (a), Osteolife™ supplemented with 10 % human serum (b), and MEM with 10 % human serum (c) reveals that serum promotes the deposition of minerals. Scale bars: 100 µm. Semiquantitative analysis of bound Alizarin S shows that addition of serum causes a significant (**p < 0.01) 10.3 fold increase of mineralisation (d). Alkaline phosphatase (AP) staining shows no AP activity in cells cultured with Osteolife™, whereas AP activity is detectable in cells cultured in Osteolife™ with serum and in cells cultured with MEM and serum (e-g overviews, h-j photomicrographs – Scale bars: 150 µm). Quantitative RT- PCR shows that Runx2 (k), BSP1 (l) and PTH1R (m) transcripts showed a trend of upregulation in cells cultured in Osteolife™ supplemented with 10 % human serum. vessel leakage by adding increasing amounts of human serum triggered only a small population of tendon cells to serum to the culture medium. We found that serum seems to deposit calcium crystals, addition of serum substantially play a fundamental role in proliferation and differentiation increased the percentage of cells containing calcium of human tendon cells and their expression of matrix- deposits. This finding indicates that serum increases the forming and matrix-degrading proteins. Under adipogenic susceptibility of tendon-derived cells to osteogenic stimuli. culture conditions, serum promoted the formation of In contrast, successful osteogenic differentiation of MSCs cytoplasmic lipid droplets and significantly increased the cultured in Osteolife™ does not require the presence of expression of the ligand-activated transcription factor serum (Bortolotti et al., 2015; Uzer et al., 2013). PPARγ, which is the master regulator of adipogenesis Interestingly, differentiation of tendon-derived cells (Rosen and Spiegelman, 2001). In contrast, Adipolife™ into the chondrogenic lineage did not seem to be serum- culture medium without serum did not induce adipogenesis. dependent. Aggrecan, the cartilage-specific proteoglycan From these results, we conclude that tendon cells need core protein, appeared to be expressed in cells cultured in both a stimulus directing the cells towards an adipogenic Chondrolife™ medium without serum supplementation. differentiation programme and the presence of serum for Expression of collagen II, the predominant collagen type final maturation into fat cells. While Osteolife™ without in articular cartilage, was induced by Chondrolife™ in both

www.ecmjournal.org 307

Fig. 10. Representative Western blots and zymograms of human tendon cell lysates showing MMP2, MMP9, and Pro-collagen1a1 expression of human tendon cells in dependence of different serum concentrations (0, 1, 5 and 10 %). ß-actin serves as loading control (a). Corresponding densitometric quantification of expression levels of MMP2 b( ), MMP9 (c) and Pro-collagen1a1 (d) from three different experiments are normalised to ß-actin. the absence and presence of serum. The same applied for expression of MMP2 and MMP9 (Alfredson et proteoglycans evidenced by Alcian blue staining. al., 2003; Robertson et al., 2012). Our results Future studies will have to address the potential role demonstrated that serum increases the expression of the described vascular barrier in the diseased tendon. of the ECM degrading matrix metalloproteinases This is tempting, as there are many parallels between our MMP2 and MMP9. Pro-collagen1a1, the major matrix in vitro observations and symptoms known from diseased constituting protein, does not show comparable serum tendons: dependence. Pro-collagen1a1 levels appear similar 1. Tissue turnover and cell proliferation are very limited in in tendon cell cultures supplemented with 1 or 10 % healthy tendons (Heinemeier et al., 2013), in diseased human serum. Apparently, tendon cells do not require tendons hypercellularity and hyperproliferation are high serum concentrations since they are capable of common features (Rolf et al., 2001). In our work, producing collagens also under low serum conditions we show that serum strongly affects tendon cell (Fig. 10d). proliferation. Tendon cells hardly proliferate when cultured without serum. Upon addition of serum, cells To underpin a putative role of an impaired barrier in immediately started to divide (Fig. 6g). tendon disorders, the permeability status and TJ expression 2. In tendinopathic or tenotomised tendons, calcification of blood vessels in compromised (e.g. diabetic) tendons and ossification processes, mucoid and fatty will be in the focus of our future studies. infiltrations have been observed (Buck et al., 2009; If tendon degenerative processes are a consequence Mavrikakis et al., 1989; McClure, 1983; Oliva et of, or at least abated by, increased permeability of leaky al., 2012), the underlying mechanism for it being tendon vessels, the application of drugs leading to a barrier largely unknown. In this work, we show that serum tightening might be a feasible strategy to slow down or contact is crucial for tendon-resident cells to undergo halt this detrimental progression. Therefore, restoration of differentiation to the osteoblast and adipocyte lineage. a functional barrier and consequently the niche may stand 3. Remodelling processes, occurring in tendinopathic at the basis for promising regeneration. and ruptured tendons, correspond to an increased

www.ecmjournal.org 308 C Lehner et al. The blood-tendon barrier

Conclusion Bi Y, Ehirchiou D, Kilts TM, Inkson CA, Embree MC, Sonoyama W, Li L, Leet AI, Seo BM, Zhang L, Shi S, We provide evidence for a TJ-based blood-tendon barrier Young MF (2007) Identification of tendon stem/progenitor restricting the passage of molecules larger than 10 kD. This cells and the role of the extracellular matrix in their niche. barrier contributes to the niche in which tendon stem cells Nat Med 13: 1219-1227. reside and differentiate and may be a novel therapeutic Bortolotti F, Ukovich L, Razban V, Martinelli V, Ruozi target in tendon regeneration strategies. G, Pelos B, Dore F, Giacca M, Zacchigna S (2015) In vivo therapeutic potential of mesenchymal stromal cells depends on the source and the isolation procedure. Stem Acknowledgements Cell Reports 4: 332-339. Buck FM, Grehn H, Hilbe M, Pfirrmann CW, Manzanell We would like to thank Prof. Dr. Wolf-Dietrich Krautgartner S, Hodler J (2009) Degeneration of the long biceps tendon: and Ing. Michaela Klappacher from the Department of comparison of MRI with gross anatomy and histology. Am Organismic Biology of the University of Salzburg for J Roentgenol 193: 1367-1375. technical support and for providing access to the electron Dimri GP, Lee X, Basile G, Acosta M, Scott G, microscope facility. Moreover, we acknowledge Dr. Katja Roskelley C, Medrano EE, Linskens M, Rubelj I, Pereira- Heinemeier and Dr. Peter Schjerling from the Institute Smith O (1995) A biomarker that identifies senescent of Sports Medicine of the University of Copenhagen for human cells in culture and in aging skin in vivo. Proc Natl their collaboration and scientific input. The authors would Acad Sci USA 92: 9363-9367. like to acknowledge funding from the Paracelsus Medical Fenwick SA, Hazleman BL, Riley GP (2002) The University (project R-10/05/022-LEH), the Herrmann vasculature and its role in the damaged and healing tendon. and Marianne Straniak Foundation (Sarnen, Switzerland) Arthritis Res 4: 252-260. and the Lorenz Böhler Funds (Vienna, Austria). We wish Garbuzova-Davis S, Mirtyl S, Sallot SA, Hernandez- to confirm that there are no known conflicts of interest Ontiveros DG, Haller E, Sanberg PR (2013) Blood-brain associated with this publication and there has been no barrier impairment in MPS III patients. BMC Neurol 13: significant financial support for this work that could have 174. influenced its outcome. Hagberg L, Heinegard D, Ohlsson K (1992) The contents of macromolecule solutes in flexor tendon sheath fluid and their relation to synovial fluid. A quantitative analysis. J Hand Surg Br 17: 167-171. References Heinemeier KM, Schjerling P, Heinemeier J, Magnusson SP, Kjaer M (2013) Lack of tissue renewal in human adult Abbott NJ, Patabendige AA, Dolman DE, Yusof SR, Achilles tendon is revealed by nuclear bomb (14)C. FASEB Begley DJ (2010) Structure and function of the blood-brain J 27: 2074-2079. barrier. Neurobiol Dis 37: 13-25. Hooper G, Davies R, Tothill P (1984) Blood flow and Ahmed IM, Lagopoulos M, McConnell P, Soames RW, clearance in tendons. Studies with dogs. J Bone Joint Surg Sefton GK (1998) Blood supply of the Achilles tendon. J Br 66: 441-443. Orthop Res 16: 591-596. Jones M, Clayton E, Noel C, Ladhani K, Grobbelaar Alfredson H, Lorentzon M, Backman S, Backman A, A (2003) Immunohistochemical endothelial localisation Lerner UH (2003) cDNA-arrays and real-time quantitative – a novel method of vessel delineation in tendon tissue. J PCR techniques in the investigation of chronic Achilles Immunol Methods 280: 175-181. tendinosis. J Orthop Res 21: 970-975. Kannus P (2000) Structure of the tendon connective Andersson G, Backman LJ, Scott A, Lorentzon R, tissue. Scand J Med Sci Sports 10: 312-320. Forsgren S, Danielson P (2011) Substance P accelerates Katenkamp D, Stiller D, Schulze E (1976) Ultrastructural hypercellularity and angiogenesis in tendon tissue and cytology of regenerating tendon – an experimental study. enhances paratendinitis in response to Achilles tendon Exp Pathol 12: 25-37. overuse in a tendinopathy model. Br J Sports Med 45: Klaassen I, Hughes JM, Vogels IM, Schalkwijk CG, Van 1017-1022. Noorden CJ, Schlingemann RO (2009) Altered expression Astrom M, Rausing A (1995) Chronic Achilles of genes related to blood-retina barrier disruption in tendinopathy. A survey of surgical and histopathologic streptozotocin-induced diabetes. Exp Eye Res 89: 4-15. findings. Clin Orthop Relat Res316: 151-164. Krizbai IA, Bauer H, Amberger A, Hennig B, Szabó Bauer HC, Traweger A, Zweimueller-Mayer J, Lehner H, Fuchs R, Bauer HC (2000) Growth factor-induced C, Tempfer H, Krizbai I, Wilhelm I, Bauer H (2011) New morphological, physiological and molecular characteristics aspects of the molecular constituents of tissue barriers. J in cerebral endothelial cells. Eur J Cell Biol 79: 594-600. Neural Transm 118: 7-21. Legerlotz K, Dorn J, Richter J, Rausch M, Leupin Bauer HC, Krizbai IA, Bauer H, Traweger A (2014) O (2014) Age-dependent regulation of tendon crimp “You Shall Not Pass”-tight junctions of the blood brain structure, cell length and gap width with strain. Acta barrier. Front Neurosci 8: 392. Biomater 10: 4447-4455. Bazzoni G (2006) Endothelial tight junctions: Lui PP (2013) Identity of tendon stem cells – how much permeable barriers of the vessel wall. Thromb Haemost do we know? J Cell Mol Med 17: 55-64. 95: 36-42.

www.ecmjournal.org 309

Lundborg G, Holm S, Myrhage R (1980) The role of the Rui YF, Lui PP, Chan LS, Chan KM, Fu SC, Li G synovial fluid and tendon sheath for flexor tendon nutrition. (2011a) Does erroneous differentiation of tendon-derived An experimental tracer study on diffusional pathways in stem cells contribute to the pathogenesis of calcifying dogs. Scand J Plast Reconstr Surg 14: 99-107. tendinopathy? Chin Med J 124: 606-610. Luo Y, Xiao W, Zhu X, Mao Y, Liu X, Chen X, Huang Rui YF, Lui PP, Ni M, Chan LS, Lee YW, Chan KM J, Tang S, Rizzolo LJ (2011) Differential expression (2011b) Mechanical loading increased BMP-2 expression of claudins in retinas during normal development and which promoted osteogenic differentiation of tendon- the angiogenesis of oxygen-induced retinopathy. Invest derived stem cells. J Orthop Res 29: 390-396. Ophthalmol Vis Sci 52: 7556-7564. Schrade A, Sade H, Couraud PO, Romero IA, Weksler Manske PR, Lesker PA (1985) Flexor tendon nutrition. BB, Niewoehner J (2012) Expression and localization Hand Clin 1: 13-24. of claudins-3 and -12 in transformed human brain Masaki H, Atsumi T, Sakurai H (1995) Protective endothelium. Fluids Barriers CNS 9: 6. activity of hamamelitannin on cell damage of murine skin Tempfer H, Traweger A (2015) Tendon vasculature in fibroblasts induced by UVB irradiation. J Dermatol Sci10: health and disease. Front Physiol 6: 330. 25-34. Tempfer H, Wagner A, Gehwolf R, Lehner C, Tauber Mavrikakis ME, Drimis S, Kontoyannis DA, Rasidakis M, Resch H, Bauer HC (2009) Perivascular cells of the A, Moulopoulou ES, Kontoyannis S (1989) Calcific supraspinatus tendon express both tendon- and stem cell- shoulder periarthritis (tendinitis) in adult onset diabetes related markers. Histochem Cell Biol 131: 733-741. mellitus: a controlled study. Ann Rheum Dis 48: 211-214. Tong WY, Shen W, Yeung CW, Zhao Y, Cheng SH, Mazzocca AD, McCarthy MB, Chowaniec DM, Chu PK, Chan D, Chan GC, Cheung KM, Yeung KW, Dugdale EM, Hansen D, Cote MP, Bradley JP, Romeo Lam YW (2012) Functional replication of the tendon tissue AA, Arciero RA, Beitzel K (2012) The positive effects of microenvironment by a bioimprinted substrate and the different platelet-rich plasma methods on human muscle, support of tenocytic differentiation of mesenchymal stem bone, and tendon cells. Am J Sports Med 40: 1742-1749. cells. Biomaterials 33: 7686-7698. McClure J (1983) The effect of diphosphonates on Traweger A, Lehner C, Farkas A, Krizbai IA, Tempfer heterotopic ossification in regenerating Achilles tendon H, Klement E, Guenther B, Bauer HC, Bauer H (2008) of the mouse. J Pathol 139: 419-430. Nuclear Zonula occludens-2 alters gene expression and Milatz S, Krug SM, Rosenthal R, Günzel D, Müller junctional stability in epithelial and endothelial cells. D, Schulzke JD, Amasheh S, Fromm M (2010) claudin-3 Differentiation76: 99-106. acts as a sealing component of the tight junction for ions Uzer G, Pongkitwitoon S, Ete Chan M, Judex S (2013) of either charge and uncharged solutes. Biochim Biophys Vibration induced osteogenic commitment of mesenchymal Acta 1798: 2048-2057. stem cells is enhanced by cytoskeletal remodeling but not Mizutani K, Ito M, Nakano T, Kamiya H, Sakurai M fluid shear. J Biomech46: 2296-2302. (1995) Impaired expression of interleukin 2 receptor and Vandesompele J, De Preter K, Pattyn F, Poppe B, CD45RO antigen on lymphocytes from children with acute Van Roy N, De Paepe A, Speleman F (2002) Accurate lymphoblastic leukemia in response to cytomegalovirus normalization of real-time quantitative RT-PCR data by and varicella-zoster virus. Clin Diagn Lab Immunol 2: geometric averaging of multiple internal control genes. 381-384. Genome Biol 3: RESEARCH0034. Nitta T, Hata M, Gotoh S, Seo Y, Sasaki H, Hashimoto Wallow IH, Geldner PS (1980) Endothelial fenestrae in N, Furuse M, Tsukita S (2003) Size-selective loosening of proliferative diabetic retinopathy. Invest Ophthalmol Vis the blood-brain barrier in claudin-5-deficient mice. J Cell Sci 19: 1176-1183. Biol 161: 653-660. Wang B, Liu W, Zhang Y, Jiang Y, Zhang WJ, Zhou Oliva F, Via AG, Maffulli N (2012) Physiopathology G, Cui L, Cao Y (2008) Engineering of extensor tendon of intratendinous calcific deposition. BMC Med10: 95. complex by an ex vivo approach. Biomaterials 29: 2954- Pfeiffer F, Schafer J, Lyck R, Makrides V, Brunner S, 2961. Schaeren-Wiemers N, Deutsch U, Engelhardt B (2011) Wang X, Qiu Y, Triffitt J, Carr A, Xia Z, Sabokbar A Claudin-1 induced sealing of blood-brain barrier tight (2012) Proliferation and differentiation of human tenocytes junctions ameliorates chronic experimental autoimmune in response to platelet rich plasma: an in vitro and in vivo encephalomyelitis. Acta Neuropathol 122: 601-614. study. J Orthop Res 30: 982-990. Robertson CM, Chen CT, Shindle MK, Cordasco FA, Weidman KA, Simonet WT, Wood MB, Cooney WP Rodeo SA, Warren RF (2012) Failed healing of rotator cuff 3rd, Ilstrup DM (1984) Quantification of regional blood repair correlates with altered collagenase and gelatinase in flow to canine flexor tendons. J Orthop Res2: 257-261. supraspinatus and subscapularis tendons. Am J Sports Med Wolburg H, Wolburg-Buchholz K, Kraus J, Rascher- 40: 1993-2001. Eggstein G, Liebner S, Hamm S, Duffner F, Grote EH, Rolf CG, Fu BS, Pau A, Wang W, Chan B (2001) Risau W, Engelhardt B (2003) Localization of claudin-3 Increased cell proliferation and associated expression of in tight junctions of the blood-brain barrier is selectively PDGFRbeta causing hypercellularity in patellar tendinosis. lost during experimental autoimmune encephalomyelitis Rheumatology 40: 256-261. and human glioblastoma multiforme. Acta Neuropathol Rosen ED, Spiegelman BM (2001) PPARgamma : a 105: 586-592. nuclear regulator of metabolism, differentiation, and cell Xu H, Dawson R, Crane IJ, Liversidge J (2005) growth. J Biol Chem 276: 37731-37734. Leukocyte diapedesis in vivo induces transient loss of

www.ecmjournal.org 310 C Lehner et al. The blood-tendon barrier tight junction protein at the blood-retina barrier. Invest size reasons no option for tracer perfusion. One approach Ophthalmol Vis Sci 46: 2487-2494. to test our hypothesis could be to make the barrier leaky Zhang J, Wang JH (2013) Human tendon stem by mannitol injection and look whether a tendinopathy cells better maintain their stemness in hypoxic culture develops. conditions. PLoS One 8: e61424. Zhang ZZ, Zhong SZ, Sun B, Ho GT (1990) Blood Reviewer I: How does the concept of a tendon blood supply of the flexor digital tendon in the hand and its barrier fit in with current tendinopathy treatment methods? clinical significance. Surg Radiol Anat12: 113-117. On the one hand, sclerosing agents are used to destroy Zhang J, Piontek J, Wolburg H, Piehl C, Liss M, Otten neovessels, whereas on the other hand, platelet rich plasma C, Christ A, Willnow TE, Blasig IE, Abdelilah-Seyfried (PRP) is injected into the tendon to promote healing. S (2010a) Establishment of a neuroepithelial barrier by Authors: The concept of a blood-tendon barrier fits in claudin5a is essential for zebrafish brain ventricular lumen well with the reported efficacy of sclerosing agents shown expansion. Proc Natl Acad Sci USA 107: 1425-1430. to destroy neovessels. Reduction of leaky neovessels Zhang Y, Wang B, Zhang WJ, Zhou G, Cao Y, Liu reduces extravasation of serum into the surrounding tissue W (2010b) Enhanced proliferation capacity of porcine and consequently, in line with our in vitro observations, tenocytes in low O2 tension culture. Biotechnol Lett 32: proliferation and erroneous differentiation of tendon cells. 181-187. Interestingly, methods shown to be beneficial for treatment of tendinopathies at high levels of evidence, such as eccentric loading, reduce tendon vascularity. Injection of Discussion with Reviewers platelet rich plasma into the tendon is believed to promote cell proliferation and angiogenesis by delivery of various Reviewer I: The potential involvement of the tendon- growth factors such as vascular endothelial growth factor blood barrier in the development of tendinopathies is an and is therefore regarded to promote the healing process. interesting idea, which does indeed lend itself to be tested But so far there is no convincing evidence from studies with in an animal model. Particularly as to date there are no high levels of evidence that treatment of diseased tendons valid tendinopathy animal models out there, what would with PRP is superior to control treatments. The findings your approach be to translate/transfer that idea into an presented in this paper may shed new light on potential animal model? mechanisms of PRP on tendon tissue. Authors: We agree with the reviewer regarding the notion that there is no valid tendinopathy small animal model available. If there was, we would tracer inject such animals and look whether the tracer leaks out of the vessels. Editor‘s Note: Scientific Editor in charge of the paper: Naturally occurring tendinopathies in horses are due to Juerg Gasser.